Event-detector comparison workflow

This article is the Python counterpart of the corresponding gp3tools workflow. It uses bundled synthetic data so it can be run without private participant exports.

import gp3tools as gp3

data = gp3.load_example_master()
# detect_gazepoint_fixations_ivt is available as gp3.detect_gazepoint_fixations_ivt(...)
# compare_gazepoint_event_detectors is available as gp3.compare_gazepoint_event_detectors(...)
# summarise_gazepoint_event_detector_agreement is available as gp3.summarise_gazepoint_event_detector_agreement(...)
# benchmark_gazepoint_event_detectors is available as gp3.benchmark_gazepoint_event_detectors(...)

Recommended workflow

1. Inspect column availability and create/validate a master table.
2. Run the relevant quality gates before transformation or modelling.
3. Keep preprocessing and exclusion decisions explicit.
4. Save derived tables/plots with package/version metadata.
5. For backend-adapted statistical functions, report the Python backend and validate the inferential target against the R reference when confirmatory equivalence matters.

Reproducibility checkpoint

import gp3tools as gp3

print(gp3.__version__)
print(gp3.api_status().query("r_export in @funcs"))